Backlink quality: compare publishers with evidence
By StarBacklinkPublisher selection5 min readUpdated
Evaluate the actual page, reader fit, metrics and terms. Compare candidates in an editable review matrix without turning missing evidence into a quality score.
Compare publisher evidence
This worksheet stays in your browser. It does not fetch websites, verify claims, buy links or save to your account. Download JSON to keep your work.
Set the destination, reader goal and market. Add up to eight candidates, record evidence for each area, then make your own decision. Counts describe your records, not link quality or ranking potential.
A useful backlink connects a specific reader need to a destination that answers it. Before comparing publishers, write that need in one sentence and choose the exact target page. Then ask what the source reader gains by following the link. Topic labels, a familiar domain and a large authority number are insufficient answers. Record language, intended market and the evidence you actually have about the audience; a catalog country is not a census of visitors. A placement may be appropriate for a small specialist audience even when a provider has little data. Missing data remains a research gap, not a negative verdict.
Inspect the page and editorial context
Request the actual page or a clearly specified publication format. Read nearby material and inspect how the proposed sentence and anchor fit the argument. Could a visitor find the link, understand its destination and benefit from leaving the page at that point? Compare a worked tutorial, a resource citation and a repeated template separately. Review a sample of recent editorial work and record who wrote it, what original information it offers and how corrections are handled. Google describes crawlable anchors with href and useful surrounding context; our checklist adds your concrete page evidence. A sitewide metric does not substitute for that review.
Keep metric name, provider, scope, date and period together. Ahrefs DR describes a relative domain link profile; Semrush Authority Score uses its own combined model. They are not interchangeable measurements. A value without provenance should not be used to settle a tie. Distinguish a domain estimate from data about the proposed page, and an estimated traffic figure from observed readership. The catalog's DA and PA fields are supplier-provided; its synchronization date is not automatically the date those scores were measured. Saved analytics can be historical. Do not relabel a missing or failed report as zero, average unrelated scores, or calculate an invented pass threshold.
Count referring pages and publisher domains separately. Fifty footer placements on one domain are fifty proposed page locations, not fifty independent editorial sources. Two listings or subdomains may also resolve to one registrable domain. The worksheet groups such candidates to make duplication visible; this is not an ownership audit. Different domains may share an operator, and hosted platforms may contain distinct sites. Record ownership evidence separately when independence matters. Neither a repeated template nor a high link count establishes value by itself. Compare the reader experience and what the commercial offer actually delivers before treating volume as breadth.
Review commercial conditions and link qualification
Write down price and currency, duration, renewal, exact placement, attributes, permitted edits, removal and the issue process. Leave unanswered terms explicit. Google treats ranking manipulation through bought links as link spam; paid advertising links should be qualified with sponsored or nofollow. Do not promise that a paid placement transfers ranking credit or stays live permanently. The catalog helps compare available inventory but does not certify every page's editorial quality. Publisher identities for new purchases remain hidden at zero balance; customers can still view publishers for their own paid links. A shortlist is a research decision, not an order or a supplier guarantee.
The editable matrix compares up to eight candidates across six evidence areas. A supported assessment or concern requires a checked date, reference and note; unknown may retain a question without becoming a pass. Shortlisting or declining requires a reviewer, date and reason. You can shortlist while explicitly accepting an unresolved measurement gap, so the counters never decide for you. Editing a candidate or its evidence clears the earlier decision for review. Context locks once a candidate is saved to avoid reusing evidence for a different purpose. Three fictional examples contrast a relevant tutorial with missing metrics, a high-metric unrelated portal, and a repeated footer from the first domain. None is a real publisher recommendation.
Evidence checklist
Review area
Evidence reference / method
Audience fit
Name the reader need, page topic, language and market. Explain why visiting your destination would help.
Editorial evidence
Inspect recent work, authorship, original information and who takes responsibility for corrections.
Placement and qualification
Record the exact location, surrounding sentence, anchor, visibility and agreed paid-link qualification.
Metrics and provenance
Name the provider, metric, page/domain scope, observation date and period. Missing measurements are not zero.
Technical evidence
Record source access, exact href, redirects and observed directives separately. A found link does not prove indexing.
Terms and maintenance
Record price, currency, duration, changes, removal, issue process and the person responsible.
Save JSON to resume the review or hand over its complete records. CSV exports one row per evidence area; the offline HTML report includes the context, decisions and unknowns. These browser-only files do not synchronize to your account, inspect the websites or place orders. After publication, compare the actual source page, href, anchor and agreed attributes with the brief, then use the monitor for a known source/target pair. Keep publication checks distinct from indexing and business outcomes. If evidence changes, retain the old export and record a fresh decision rather than quietly treating the earlier shortlist as a current endorsement.
No. Keep the provider, scope and date, then inspect the actual page and reader fit. This worksheet does not create a pass score.
Can a high metric guarantee a good placement?
No. A domain measurement cannot supply missing page context, audience evidence or commercial terms. The fictional portal demonstrates that distinction.
Do fifty sitewide links mean fifty publishers?
No. Keep referring-page counts separate from publisher domains. Domain grouping also does not prove independent ownership.
Does the worksheet verify a publisher or submit an order?
No. You enter the evidence and decisions. It saves downloadable records locally; it does not fetch, certify, buy or predict rankings.
Enough for ten standard 1-credit links for one month. Higher-priced sites or longer durations use more credits. One claim per verified customer, while campaign places remain. No automatic renewal or ranking guarantee.